Connection issues after upgrading to 0.13

I have a net with 1 debian (A) and 2 windows (B,C) machines in it. Some words about configuration:

  1. A is behind firewall with manually forwarded port, NAT traversal and global discovery disabled. Address is set as http://ip:port in other two.

  2. B is behind firewall, NAT traversal (UPnP) and global discovery enabled. Address is set as dynamic in other two.

  3. C is behind firewall, NAT traversal (UPnP) disabled and global discovery enabled. Address is set as dynamic in other two.

There are some connection issues after upgrading to 0.13 which weren’t observed on 0.12:

  1. A sees B and C. No problem here.

  2. B sees A and doesn’t see C.

  3. C sees A and doesn’t see A.

I have to restart B and C several times and with the touch of luck they may connect.

UPDATE: As far as I can get they connect after all but it takes some time. At least today they connected after 17 minutes uptime of node C.

There was a bug that was fixed in 0.13.1, plus if you use your own discovery you need to upgrade discovery server.

If it connects directly after some time, it’s likely to be related to discovery caching non-existance of the device.

If it connects via a relay (shows connection type relay in the details section for the device), then it can’t connect directly most likely.

Ok, thanks for the reply, let’s hope the problem is gone. All nodes are updated to 0.13.1. Secondly, I don’t use neither own discovery server nor relays.

If you did not change the default config (relay enabled) Syncthing will use the default relay pool provided by Syncthing / the community. So it is still possible, that they connect through relay, because direct connection isn’t possible.

Do you mean this?

Am I right that when relay is enabled all data is transferred from one device to another through provided relay server? Or is it just used to establish connection between devices?

It uses a relay if direct connection is not available.

1 Like

If relay is enabled and the devices cannot connect directly to one another (see https://docs.syncthing.net/users/firewall.html) they will use the relay to connect and transfer data. Transferring through public relays is slow (normally capped at 100KiB/s). The data is encrypted, so that the relay cannot see the actual data.

If relays is disabled, they will just not be able to connect.

1 Like

If relays is disabled, they will just not be able to connect.

I see them connecting quite well now even with disabled relays. Isn’t this true anymore?

If both sides are behind NAT, one side needs a port forward, either manually or by UPnP.

I guess that’s exactly my case with B and C.

If relays is disabled, they will just not be able to connect.

This was still in regards to the assumption:

… and the devices cannot connect directly to one another

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.